Chinese artificial intelligence company Moonshot AI announced the launch of its latest AI model, Kimi K3, on Friday. The company stated that this new model, which contains 2.8 trillion parameters, now represents the most extensive open-weight AI system available globally and offers performance levels nearing those of leading U.S.-developed models.
Moonshot’s unveiling comes shortly after the U.S. government ordered the removal of Anthropic’s Fable and Mythos models in May, citing security issues. The roll out of Kimi K3 highlights China’s rapid progress in developing open AI technologies that are increasingly rivaling advanced American systems.
According to Moonshot, while Kimi K3 still falls short of matching the capabilities of Anthropic’s Claude Fable 5 and OpenAI’s GPT 5.6 Sol, it delivered stronger results than models such as Claude Opus 4.8 and GPT 5.5 on a variety of benchmarks, including those measuring coding proficiency and task management.
The model was built to perform sophisticated reasoning and coding over extended sessions, with Moonshot indicating Kimi K3 is the first open-weight model to approach the 3 trillion-parameter threshold. Its 1 million-token context window allows it to process and hold more information per prompt than earlier models.
Open-weight models, unlike proprietary closed-source alternatives, can be freely downloaded, operated, and modified by users. Parameter count—while not a definitive indicator of a model’s abilities—remains a commonly cited metric for size. Direct evaluation against top U.S. systems is complicated as leading American developers, such as OpenAI and Anthropic, do not disclose parameter figures for their newest models.
Chinese companies such as Moonshot, Z.ai, and MiniMax have been accelerating innovation, bringing powerful and cost-efficient models to the market and challenging the perception that Chinese developers lag Silicon Valley by months.
With major investors including Alibaba and Tencent, Moonshot has been ramping up resources and technological development to maintain a leading position in the competitive AI sector. The growing use of Chinese AI models in Western businesses, coupled with questions among U.S. lawmakers about restricting their adoption, reflects the intensifying race between the two countries.
